feature selection or feature reduction technique are PCA, Factor analysis.Nearest neighbor search

MinHash

Information gain in decision trees

Semidefinite embedding

Multifactor dimensionality reduction

Multilinear subspace learning

Multilinear PCA

Singular value decomposition

Latent semantic analysis

Semantic mapping

Topological data analysis

Locality sensitive hashing

Sufficient dimension reduction

with svm i go with multilinear PCA and achieve more then 90% accuracy. Majority case SVM is better then ANN

The optimal method to get the best features out of n is to apply exhaustive search, that is, you need to test all possible combinations of the features (nth power of 2). However, the computational time would be drastically high for even moderate numbers of the features. As a suboptimal solution, you may prefer filter- or wrapper-based feature selection approaches.

An appropiate complexity level can also improve the efficiency: number of hidden units in the network, etc

There are a bunch of methods to perform this in the machine learning community, as said before you can look at feature selection approaches, or at learning methods such as SVM or regression where you add a regularization term that gives you a sparse model. So keywords like "sparsity" and "feature selection" can lead you to possible methods.

You will find basic methods in textbooks such as "pattern recognition" or "machine learning - a probabilistic perspective" and the like.

However the first thing to do is probably to run some basic statistical test/method (scatter plot, etc...) to test the dependence between your features and the output, before trying any fancy method.

Hi..

Feature Selection is basically selection of a subset of features from a larger pool of

available features. The goal is to select those that are rich in discriminatory information with respect to the classification problem at hand. This is a crucial step in the design of any classification system, as a poor choice of features drives the classifier to perform badly. Selecting highly informative features is an attempt

(a) to place classes in the feature space far apart from each other (large between-class distance) and

(b) to position the data points within each class close to each other (small within-class variance).

For more details: see chapter-4 of below book:

Book: An introduction to pattern recognition: A MATLAB approach

BY: Theodoridis, Sergios

You can also  try WEKA data mining tool for feature selection, as in WEKA different feature selection algorithms are there.

Gud luck...!!!!